Tesla Model X

GRA said:

I don't think there's any doubt that the severity of the damage was due to the out-of-action barrier. The far more important question as far as the investigation goes is did hitting the barrier result from human or A/P error?

Click to expand...

Data represents whether statistically-wise new car features
... (seat-belts, A.E.Braking, Lane-Keep-Assist, etc from many companies and auto-pilot from Tesla or variations from other companies) ...
... are saving more lives than not having these features.

scottf200 said:

GRA said:

I don't think there's any doubt that the severity of the damage was due to the out-of-action barrier. The far more important question as far as the investigation goes is did hitting the barrier result from human or A/P error?

Click to expand...

Data represents whether statistically-wise new car features
... (seat-belts, A.E.Braking, Lane-Keep-Assist, etc from many companies and auto-pilot from Tesla or variations from other companies) ...
... are saving more lives than not having these features.

Click to expand...

Yeah, but how many of those are under the same lighting and sun conditions as the driver that died? How many were in the same lanes and how many required the driver to intervene? And, we don't know if this driver used the auto-lane change feature and got changed into a non-lane w/the concrete barrier.

We also don't know how clean the camera lenses were and if all cameras were working on this particular X. The software revision of all those vehicles is likely to be not all the same.

I haven't had time due to other priorities to follow that monster thread but https://teslamotorsclub.com/tmc/threads/model-x-crash-on-us-101-mountain-view-ca.111505/page-28#post-2642486 is supposedly what the lighting and sun looked like right before the crash.

scottf200 said:

GRA said:

I don't think there's any doubt that the severity of the damage was due to the out-of-action barrier. The far more important question as far as the investigation goes is did hitting the barrier result from human or A/P error?

Click to expand...

Data represents whether statistically-wise new car features
... (seat-belts, A.E.Braking, Lane-Keep-Assist, etc from many companies and auto-pilot from Tesla or variations from other companies) ...
... are saving more lives than not having these features.

Click to expand...

I sure the data will also show 100 million successful trips along that route every year along with a dozen deaths. I guess we good then, right?

The reality is we have nearly perfected the art of self sacrifice despite the best efforts of manufacturers to prevent it.

DaveinOlyWA said:

I sure the data will also show 100 million successful trips along that route every year along with a dozen deaths. I guess we good then, right? The reality is we have nearly perfected the art of self sacrifice despite the best efforts of manufacturers to prevent it.

Click to expand...

I'm confused.

There are many 1000s of car accidents daily with people getting hurt to varying degrees and even dying in normal all manually driven cars. Right?

The goal is to lessen the accidents, deaths, family grief, and all surrounding things related to this (insurance cost rising, worker (police, fire, city road), cost, cleanup cost, rebuilding safety road features guards, barrels, compression, etc).

^^^Sure, that's the goal, but without public buy-in AVs won't happen. People's concerns about turning their safety over to a computer are inevitable, regardless of whether or not doing so is statistically safer. AVs are supposed and must be able to prevent this sort of all-too-common human-caused accident. If they can't handle routine problems like this in good conditions, how are they supposed to handle corner cases in bad ones?

As of yet, we simply don't know what the cause of this accident was. Depending on the amount of damage suffered to the car's memory, we may never know, but it is inevitable that early fatal self-driving accidents will attract a whole lot of attention, regardless of what the statistics say (and Elon was called out by statisticians after Joshua Brown's accident for making safety claims for A/P that violated all sorts of statistical methodologies). We'll just have to wait and see what NTSB says - if this was a human error, very sad, but we're used to those and move on. If not, Tesla's on the hook for a bunch of money and takes a major PR hit, as do AVs, but it's the Uber crash that will really stoke public resistance. After all, if you're riding in an AV you've made that choice for yourself, but if you get hit by one you haven't.

GRA said:

The far more important question as far as the investigation goes is did hitting the barrier result from human or A/P error?

Click to expand...

These stats are about THAT same stretch of HIGHWAY.
20,000 in the past 3 months.
200 per day.

Very strange assertion here:

If the driver experienced repeated AP failures in a particular location, why would he trust AP there on the day he died?

I-TEAM EXCLUSIVE: Victim who died in Tesla crash had complained about auto-pilot

...The NTSB told Dan Noyes from their Washington headquarters, they recovered both the restraint control module and infotainment module on Wednesday...

We're learning much more about the victim -- 38-year-old Walter Huang had a wife and two kids, lived in Foster City, worked for 13 years as a programmer at Electronic Arts. This past November, he got a job as an Apple engineer and bought his new Tesla, posting a picture on his Facebook page...

Dan Noyes also spoke and texted with Walter Huang's brother, Will, today. He confirmed Walter was on the way to work at Apple when he died.

He also makes a startling claim -- that before the crash, Walter complained "7-10 times the car would swivel toward that same exact barrier during auto-pilot. Walter took it into dealership addressing the issue, but they couldn't duplicate it there."

Noyes: "The family is telling me they provided an invoice to investigators, that the victim took the car in because it kept veering at the same barrier. How important is that information?"
O'Neil" "That information has been received by the CHP, they've been acting on it for some time now."

Tesla would not comment on the information we've learned...

Click to expand...

http://abc7news.com/automotive/i-team-exclusive-victim-who-died-in-tesla-crash-had-complained-about-auto-pilot/3275600/

I think TSLA's blame-the-driver strategy may not be sufficient this time...

An Update on Last Week’s Accident

The Tesla Team March 30, 2018

...In the moments before the collision, which occurred at 9:27 a.m. on Friday, March 23rd, Autopilot was engaged with the adaptive cruise control follow-distance set to minimum. The driver had received several visual and one audible hands-on warning earlier in the drive and the driver’s hands were not detected on the wheel for six seconds prior to the collision. The driver had about five seconds and 150 meters of unobstructed view of the concrete divider with the crushed crash attenuator, but the vehicle logs show that no action was taken...

Click to expand...

https://www.tesla.com/blog/update-last-week%E2%80%99s-accident

** Driver's make choices in how to use the driver assistant feature **

READ the last four paragraphs. Obviously a very sad situation for the family and everyone involved.

Here is what Tesla wrote:

Tesla said:

The driver had five seconds and 150 meters of unobstructed view of the concrete divider with the crushed crash attenuator, but the vehicle logs show that no action was taken.

Click to expand...

A more appropriate statement might have been:

Autopilot, which was controlling the car at the time, had an unobstructed view of the concrete divider yet it chose to drive directly into it, resulting in the death of the driver.

Click to expand...

And they go on to lie with statistics:

Tesla said:

In the US, there is one fatality every 86 million miles across all vehicles from all manufacturers. For Tesla, there is one fatality, including all known pedestrian fatalities, every 320 million miles in vehicles equipped with Autopilot hardware. If you are driving a Tesla equipped with with Autopilot hardware, you are 3.7 times less likely to be involved in a fatal accident.

Click to expand...

That may all be true, but the implication by the wording is that the reduction in fatality rate is somehow influenced by the availability of Autopilot hardware. There is no evidence one way or the other whether Autopilot improves Tesla's statistics. In fact, Autopilot might be making them worse than they would be otherwise.

But the simple fact is that other factors are likely much more important in these statistics. For instance, the Tesla Model S, which accounts for most of the vehicles with Autopilot included, received the best crash rating of any car ever, IIRC. Also, the Tesla Model X and Model X are heavier than most sedans on the road, so they will tend to fare better than smaller vehicles in accidents. Also note that Autopilot is only capable of driving in the simplest and safest environments. It does not take on the most challenging driving tasks that humans must handle. Finally, since Tesla ALWAYS blames the driver for fatal crashes with Autopilot engaged, if follows that Autopilot gets NO credit for miles covered when there are not crashes: the driver gets all that credit also. In other words, Autopilot has no miles of driving without fatalities since all miles with fatalities are credited to the driver.

All this in an attempt to draw attention away from the fact that Autopilot drove this vehicle directly into a concrete divider barrier, killing the driver.

RegGuheert said:

Autopilot drove this vehicle directly into a concrete divider barrier, killing the driver.

Click to expand...

The driver determined on their own to enable the driver assistance feature ** moments ** before the crash and set the following distance to 1 (out of 7).

10s of 1000s of drivers make choices every single day that cause them to get into an accident. Many 1000s every day.

https://www.teslarati.com/tesla-confirms-autopilot-engaged-model-x-fatal-crash/

Local news agency ABC7 News was able to get in touch with the driver of the vehicle that collided with the crash cushion 11 days before the Tesla accident. According to the news agency, the previous crash involved James Barboza, who was driving a 2010 Toyota Prius at 70 mph. Barboza walked away from the crash with lacerations on his face and complaints of pain all over his body. The Toyota Prius driver was eventually arrested for driving under the influence.

In a statement to ABC7, Steven Lawrence — a lawyer who specializes in highway safety — stated that the crash cushion, which could have saved the Model X driver’s life, should have been repaired long before the accident. According to Lawrence, 11 days is far too long to fix a crash cushion, especially in areas where the Model X accident took place.

“Some states have as short as a 3-day repair time for high traffic locations. And if you look at the material in California, this thing should have been repaired within a week. Again, there are a lot of questions about what happened and what went wrong, but it should have been repaired in under 11 days.” Lawrence said.

Click to expand...

An accident within 7 days would have had someone crying the barrier needed to be replaced in 3 days...

scottf200 said:

RegGuheert said:

Autopilot drove this vehicle directly into a concrete divider barrier, killing the driver.

Click to expand...

The driver determined on their own to enable the driver assistance feature ** moments ** before the crash and set the following distance to 1 (out of 7).

10s of 1000s of drivers make choices every single day that cause them to get into an accident. Many 1000s every day.

Click to expand...

The thing is, each of those drivers are liable for their own mistakes. The cost is distributed. However, if you sell an "autopilot" that accepts and executes a blatantly unsafe command, then you as a manufacturer are liable for an unsafe product. The drivers' lack of responsibility in the poor selection doesn't absolve Tesla, especially if there is harm to third parties. And it's this principle across automated driving as a whole, that will come to bite the manufacturers hard -- whether their cars are safer "on average" than a human driver or not. The costs will be concentrated, not distributed. The only way around this is for government to intervene and deny rights of victims to sue for full compensation.

ISTM that while the severity of the damage and the fatal injury are specific to the Model X, now that Tesla has confirmed that the car was under the control of A/P at the time of the crash, that discussion belongs in either the "Autonomous Vehicles, Leaf and others" or "Tesla's Autopilot, on the road" topics. It's as yet unclear whether this crash was due to some problem specific to Tesla or was more generic to AV systems. Now that companies other than Tesla have vehicles with some AV capability out in the real world and such crashes will become increasingly frequent, I think the general AV topic is the correct one, and I'll be posting any further news/comments related to the causes of this accident there, along with those relating to the Uber accident, legal fallout, etc.

Now that Tesla KNOWS that their Autopilot product drove into this concrete barricade, I think this is a perfect opportunity for Tesla to put their OTA update capability into service in order to "do the right thing" with regards to this accident. They can instruct Autopilot that this stretch of highway is a "lock-out zone" until they have been able to determine exactly why the Autopilot drove this customer to his death. Once product defect is understood, corrected, fully tested in the question area and the conditions are added to the regression test suite, then they can unblock this section of road.

Bjorn tests* bio-weapon defense mode in his Model X, via IEVS: https://insideevs.com/bjorn-stink-bombs-his-own-model-x-in-the-name-of-science/

*Requirements: a friend?, one can of odorous Swedish fish, two plastic buckets, a can opener, and a shower cap. Watch the video, it's hilarious.
Reminds of my Teamster days, when I once worked in a freight terminal (aka barn) that had to handle a container full of raw hides that had been sitting closed in the sun all day. I didn't have to work it myself (if they hadn't finished it before my turn came, I would have quit), but just walking past the back of the trailer as much as 10 feet away it was as if you physically collided with a wall of revolting scent - you recoiled. I would have been puking my guts out if I'd been inside the container, and I don't know to this day how the two unfortunates who had to work it managed.

Via IEVS:

Bjørn Descends 18% Grade In Tesla Model X Without Brakes

Click to expand...

https://insideevs.com/bjorn-descends-18-grade-in-tesla-model-x-without-brakes/

. . . Was his Model X able to pull it off? Well, there are a few tense moments when the car gained speed just before a turn. And during the short drive he suggests that the Ampera-e might be better suited to this due to it’s higher regenerative braking. . . .

Click to expand...

Of great interest to me, as a lot of my driving involves steep mountain roads, including regularly descending Old Priest Grade (18-20%) on my way back from Yosemite, and the east side of Sonora Pass is 25 or 26%. I really liked the Bolt's regen selections when I test drove it, although I didn't have a chance to descend any steep roads.

GRA said:

Via IEVS:

Bjørn Descends 18% Grade In Tesla Model X Without Brakes

Click to expand...

https://insideevs.com/bjorn-descends-18-grade-in-tesla-model-x-without-brakes/

. . . Was his Model X able to pull it off? Well, there are a few tense moments when the car gained speed just before a turn. And during the short drive he suggests that the Ampera-e might be better suited to this due to it’s higher regenerative braking. . . .

Click to expand...

Of great interest to me, as a lot of my driving involves steep mountain roads, including regularly descending Old Priest Grade (18-20%) on my way back from Yosemite, and the east side of Sonora Pass is 25 or 26%. I really liked the Bolt's regen selections when I test drove it, although I didn't have a chance to descend any steep roads.

Click to expand...

I watched the video and it seemed like the regen worked perfectly and it did not appear to be ANY tense moments at all. All those 'switch backs' had flat spots and the car slowed down easily to like 6 to 8 mph.

It would be pretty hard to touch the brake for a second or two so this is critical /sarcasm.

scottf200 said:

GRA said:

Via IEVS:

Bjørn Descends 18% Grade In Tesla Model X Without Brakes

Click to expand...

https://insideevs.com/bjorn-descends-18-grade-in-tesla-model-x-without-brakes/

. . . Was his Model X able to pull it off? Well, there are a few tense moments when the car gained speed just before a turn. And during the short drive he suggests that the Ampera-e might be better suited to this due to it’s higher regenerative braking. . . .

Click to expand...

Of great interest to me, as a lot of my driving involves steep mountain roads, including regularly descending Old Priest Grade (18-20%) on my way back from Yosemite, and the east side of Sonora Pass is 25 or 26%. I really liked the Bolt's regen selections when I test drove it, although I didn't have a chance to descend any steep roads.

Click to expand...

I watched the video and it seemed like the regen worked perfectly and it did not appear to be ANY tense moments at all. All those 'switch backs' had flat spots and the car slowed down easily to like 6 to 8 mph.

It would be pretty hard to touch the brake for a second or two so this is critical /sarcasm.

Click to expand...

As long as it's at least the equivalent decel of 2nd gear compression braking in a stick it's fine, but I prefer to have more real time control than having to choose it on the touchscreen. That's why I like the Bolt (and Volt), as you've got 'D' or 'B' on the shift lever with an easy toggle (pull back and release) between the two, and you can add extra regen with the paddle in either mode. Coming down from say Tioga Pass, I expect I'd be coasting in 'D' most of the way, using the paddle to slow me as necessary, and rarely wanting 'B' mode. It's nice to have 'B' plus paddle for those really steep downhills like Old Priest, especially when heavily loaded. If I'd learned to drive in a car with constant heavy regen it might be a different story, but as I've always driven manual trannies I prefer being able to select/adjust my regen levels on the fly, without needing to feather the accelerator to coast.